Demo: mmFHE: mmWave Sensing with End-to-End Fully Homomorphic Encryption
Year: 2026
Type: conference-paper
Abstract: In this work, we demonstrate mmFHE [2], an mmWave sensing system with end-to-end fully homomorphic encryption (FHE). Offloading the compute-heavy sensing pipeline to the cloud supports applications that resource-constrained edge devices cannot run locally, but the cloud then gets access to raw radar data that contains heartbeat, respiration, and user identity. mmFHE addresses this by encrypting radar data on the edge device and executing the entire cloud-side DSP and ML inference pipeline on ciphertexts using data-oblivious FHE kernels. The cloud returns encrypted results that only the trusted client can decrypt. This demo will showcase live wireless vital-sign monitoring with a 77 GHz FMCW radar: attendees can see their real-time heart rate and respiration rate, computed entirely on a cloud GPU server that observes only encrypted radar data.
Institution: Cornell Tech
Language: English
Subfield: Aerospace Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7223083379